[ntp:hackers] Simplifying NTPD

Loganaden Velvindron loganaden at gmail.com
Wed Jul 2 10:40:49 UTC 2014


On Tue, Jul 1, 2014 at 11:08 PM, Poul-Henning Kamp <phk at phk.freebsd.dk> wrote:
> In message <53B2EBA8.5030307 at ntp.org>, Danny Mayer writes:
>>On 6/30/2014 10:58 AM, Poul-Henning Kamp wrote:
>
>>We need to look at the overall design before making an architectural
>>decision like that. The biggest problem with things like stdout is that
>>a service has no standard output and you need to redirect it elsewhere.

Before we make the architectural changes, I would be really happy if we could
maybe push a release, and then work on those architectural changes in
the next release ?

The last release was in 2011. Since then, quite a few changes have
been implemented,
such as seccomp sandboxing.

I'm currently running ntp as snapshots on my public servers, and i
would be very happy
if I could get a binary from the distribution.



>
> We are not talking about an unrelated process, so forget about the
> local environment of services: we are talking about a process
> specifically fork+exec'ed from NTPD.
>
> NTPD will set the stdout for the child process to point the right
> place.
>
> On UNIX, it is natural to use a pipe(2).
>
> On platforms without pipe(2) any other byte stream which can be
> mapped onto stdout can be used: UNIX sockets, TCP connections etc.
>
> This is really not complicated at all -- that's the entire point.
>
> --
> Poul-Henning Kamp       | UNIX since Zilog Zeus 3.20
> phk at FreeBSD.ORG         | TCP/IP since RFC 956
> FreeBSD committer       | BSD since 4.3-tahoe
> Never attribute to malice what can adequately be explained by incompetence.
> _______________________________________________
> hackers mailing list
> hackers at lists.ntp.org
> http://lists.ntp.org/listinfo/hackers



-- 
This message is strictly personal and the opinions expressed do not
represent those of my employers, either past or present.


More information about the hackers mailing list